A key feature in most cellular communication systems is the ability to handoff an ongoing communication service from one cell to another. Handover (HO) methods and algorithms can be classified in many different ways, e.g. as soft handover where a mobile station is connected to several base stations, softer handover where a mobile station is connected to several cells or sectors belonging to the same base station, and hard handover where the mobile station disconnects from the old base station before connecting to the new base station. Methods for handover decisions can be also be classified as being network controlled HO (NCHO), in which the mobile is passive, mobile assisted HO (MAHO), in which the mobile e.g. measures the strengths of received signals and reports the measured values to the network where a handover decision is then taken, and mobile controlled HO (MCHO), in which the mobile e.g. measures the strengths of received signals and makes a handover decision based on the measured values.
One important class of handover algorithms is the radio-signal-measurement (RSM) triggered schemes. Most RSM triggered handover schemes perform averaging or low-pass filtering of measured data. Furthermore, the handover decision algorithms belonging to this class typically include, at least, a hysteresis margin and a time-to-trigger threshold that the filtered data samples are compared against during the handover decision process.
The 3rd Generation Partnership Project (3GPP) is a collaboration agreement that brings together a number of telecommunications standards bodies. Within the 3GPP workgroups a new system concept denoted Long Term Evolution (LTE) and System Architecture Evolution (SAE) are being standardized. The architecture of the 3GPP LTE/SAE system (denoted LTE here after), which is schematically illustrated in FIG. 1, is flat compared to e.g. GSM (Global System for Mobile communications) and WCDMA (Wideband Code Division Multiple Access) based systems. FIG. 1 shows that the LTE radio base stations 100a, 100b, 100c (denoted eNodeBs, or eNBs, in 3GPP terminology) are directly connected to the core network nodes 101a, 101b MME/S-GWs (mobility management entity/serving gateway) via the S1 interfaces 102a, 102b, 102c, 102d. The S1 interface supports a many-to-many relation between MMEs/Serving Gateways and eNBs. There is no central radio network controller in the Evolved UMTS Terrestrial Radio Access Network (E-UTRAN). Instead the eNBs are connected to each other via the direct logical X2 interfaces 103a, 103b, 103c. 
The handover method that will be used in the 3GPP LTE/SAE system is RSM triggered and the mobile assisted (MAHO) hard handover. In LTE the mobile station, also referred to as the user equipment (UE), performs measurements of the downlink and the network makes the handover decisions. Compared to legacy cellular systems, as stated above, the LTE system does not have any central radio network controller (like the BSC in GSM and the RNC in WCDMA) where the handover algorithm is located. Instead the handover decisions in LTE will be performed in the base stations (referred to as eNBs in LTE). The decision to initiate a handover from a source cell to a target cell will be made in the source cell by the radio base station.
The UEs are configured by a radio resource control (RRC) entity in the source cell to perform measurements on handover candidate cells and to report these measurements to the source eNB during active mode. The details of how these measurements are configured are not yet decided in 3GPP. The handover measurement configuration will either be broadcasted or sent as dedicated messages to each individual UE. Both options will probably be possible in the finalized LTE standard.
The RRC messages for configuration of handover measurements as well as the corresponding UE measurements will be standardized and will not be subject to vendor specific interpretation or implementation. A typical configuration is that the UE will start to report periodically to the radio base station of a handover candidate cell once the filtered reference symbol received power (RSRP) of the candidate has reached a certain level compared to the RSRP level of the source cell during a configurable time. Once the UE starts reporting it could continue to obtain measurements and continuously send updated reports. Alternatively, the UE could perform measurements only prior to reporting and then send a single report stating that prerequisites for a handover are fulfilled.
FIG. 2 shows a diagram that illustrates a conventional handover procedure from a source cell to a target cell in LTE. The vertical axis shows signal level and the horizontal axis shows time. The UE is configured by the source cell RRC to perform measurements on the source cell RSRP (RSRP1) and on candidate cells RSRP, i.e. possible target cells to which handover might be likely to occur. It should be noted that only one candidate cell measurement RSRP2 is shown. The measurement command contains information about how the UE shall process, e.g. by filtering or averaging, the measured data and when the UE shall start to report measurements to the source eNB. In this example the UE is configured to start to perform periodic reporting once the candidate RSRP2 value is larger than the source cell RSRP1 plus a hysteresis margin 21 during a certain time period 22 (time-to-trigger, or TTT). This occurs at a time denoted Ta. The purpose of the hysteresis margin is to prevent that action is taken prematurely. The hysteresis margin is defined as a predefined minimum difference between measurement values. In the example in FIG. 2 the hysteresis margin 21 defines a minimum difference between RSRP1 and RSRP2. After the source eNB1 has received one or several reports from the UE a decision to initiate handover to the target eNB2 is taken by eNB1. The eNB1 sends a handover request to eNB2 at a time denoted Tb, and when the handover is prepared the eNB1 sends at a time denoted Tc a handover command instructing the UE to perform the handover to eNB2.
The handover decision process in the source eNB typically consists of processing, e.g. by means of low-pass filtering, the UE measurements and by comparing the processed values with a hysteresis margin and with a time-to-trigger threshold. The eNB may use different handover related parameters when deciding on handovers to different target cells. Furthermore the eNB may classify UEs based on the speed or their handover history and the eNB may use different parameters for the handover decision process for different UE classes. In this way the eNB may e.g. use a particular set of handover related parameters for UEs that are classified as high speed UEs. For high speed UEs the time-to-trigger might need to be reduced compared to low speed UEs. Alternatively, the eNB in a first cell may know that UEs that enter from a second cell will almost always perform handover to a third cell. To ensure that such UEs, that might be moving in a train or along a road, end up in the correct target cell the hysteresis margin to the desired target cell may be reduced for this particular class of UEs.
The fact that the handover algorithm in LTE is performed by the eNB, and not in a central node controlling several base stations, as e.g. in an RNC in the WCDMA based UMTS, results in several problems that need to be addressed.
To begin with, there is no simple way to ensure that handovers within a geographical area are performed based on the same algorithm. This becomes particularly difficult in a multi vendor scenario since it is likely that different eNB vendors will implement different proprietary handover algorithms. Consequently, the criteria for when to perform a handover between two cells of the same type may be completely different depending on which cell that acts as source cell. Furthermore, within an area the criteria for when to perform a handover from an LTE system to a system having a different radio access technology (e.g. WCDMA) may also differ depending on which cell that currently is serving a particular UE.
Another problem is related to the planning and optimization of the handover related parameters that control the behaviour of the handover algorithm(s). In case of network planning the operator may be faced with the difficult task of setting a large number of handover related parameters corresponding to the particular algorithm implementations of different vendors. Each parameter will have its own definition and impact on the handover behaviour.
According to the 3GPP the handover preparation phase is initiated when the source eNodeB sends a HANDOVER REQUEST (HO_REQUEST) message to a target eNB via the X2 or the S1 interface. It has been proposed but not agreed in 3GPP to add an optional information element denoted HO_RRM_CONTAINER into the HO_REQUEST message. The content of this container is proposed not to be subject to standardization and hence an eNB vendor may put proprietary information into it. In case both the source and the target eNB are manufactured by the same vendor this optional container can be useful for support of more advanced handover methods. To have some consistency in the handover behaviour in the network the operator may then decide to only employ eNBs from one vendor in a certain area. However also this might be problematic, since in different types of base stations (macro, micro, pico) it can make sense to implement different handover algorithms. Macro-cells, micro-cells, and pico-cells, respectively, refer to cells of different sizes, whereby a macro-cell, which is a normal cell, is the largest, and a pico-cell is the smallest. For example, a pico-cell may not have to handle handover of high speed UEs.
As mentioned above, existing solutions for handover parameter optimization rely on the handover being performed in a central controller node. This is not applicable for LTE. Instead the handover decisions are distributed to the radio base station where different vendors may implement their own proprietary algorithms.
The complexity of the problem of manually optimizing handover parameters in a multi-vendor scenario is large. Handover parameter auto-tuning, or auto-adjusting, methods are complicated by the fact that they require extensive efforts regarding standardizing algorithms, eNB measurements, and parameters related to handover.
Furthermore, handover decisions are frequently made based on inadequate, and often out-of-date, data. This may lead to handovers being executed too fast, too slow or sometimes unnecessarily. This may hence lead to a non-optimal use of radio resources.
The work in 3GPP is focused on adding information in the HO_ REQUEST message sent from the source to the target eNB. However since the handover decision by then is already taken by the eNB in the source cell, the added information is of little use when it comes to auto-tuning of handover related parameters in the source cell.
Thus, optimization of handover related parameters in an architecture in which the handover decisions are made by the radio base station, as e.g. in LTE, presents a problem to which the prior art does not offer satisfactory solutions.